Load expression data

## # A tibble: 6,634 x 4
## # Groups:   Gene [6,634]
##    Gene        RNA    RPF    TE
##    <chr>     <dbl>  <dbl> <dbl>
##  1 CNM01300  3981. 18260. 4.59 
##  2 CNM01080  8422.  8957. 1.06 
##  3 CNA07570  5811.  7134. 1.23 
##  4 CNG04360  3171.  7048. 2.22 
##  5 CNB02360  3764.  6958. 1.85 
##  6 CNA06350 15019.  6591. 0.439
##  7 CNC00700  2344.  6224. 2.65 
##  8 CNF03840 11321.  6175. 0.545
##  9 CNF02150 15611.  6144. 0.394
## 10 CNF03160  5094.  6114. 1.20 
## # ... with 6,624 more rows

We also calculated hiTrans_JEC21, the top 5% (330) translated genes by RPF TPM.

Ribosome occupancy mostly tracks RNA abundance

Load context data

## # A tibble: 6,636 x 19
##    Gene  aATG.context aATG.pos d1.context d1.posTSS d1.posATG d1.frame
##    <chr> <chr>           <dbl> <chr>          <dbl>     <dbl>    <dbl>
##  1 CNA0… GACCCCCTTGT…       93 ATAGCTGGT…       226       133        1
##  2 CNA0… ATATTGCCTGA…      102 GTCCACCTT…       163        61        1
##  3 CNA0… GAACTATCAAG…      214 GAGGCTCCG…       512       298        1
##  4 CNA0… ATTTTCAACAG…       81 AGCAATATA…       307       226        1
##  5 CNA0… ACCGTGCACAC…       76 GTATTCGGG…       106        30        0
##  6 CNA0… AATCATACCAA…      117 GCCCCTATC…       186        69        0
##  7 CNA0… CCGACTATAAA…       52 AACCGTGCT…       112        60        0
##  8 CNA0… CTTTCTCTTCA…       77 TGCTATAGC…        98        21        0
##  9 CNA0… TAATCACACAA…      330 CTCATCATC…       391        61        1
## 10 CNA0… AAAAAAAACGC…      146 ACTTGTCGA…       184        38        2
## # ... with 6,626 more rows, and 12 more variables: d2.context <chr>,
## #   d2.posTSS <dbl>, d2.posATG <dbl>, d2.frame <dbl>, u1.context <chr>,
## #   u1.posTSS <dbl>, u1.posATG <dbl>, u1.frame <dbl>, u2.context <chr>,
## #   u2.posTSS <dbl>, u2.posATG <dbl>, u2.frame <dbl>

Annotated ATGs have a Kozak consensus sequence

Highly translated Annotated ATGs have a Kozak consensus sequence

That’s for hiTrans_JEC21, the top 5% (330) translated genes by RPF TPM.

Cytoplasmic Ribosome Annotated ATGs have a Kozak consensus sequence

All cytoribo genes are highly translated

Venn diagram

Upstream ATGs don’t have a consensus

First upstream ATG.

Downstream ATGs don’t have a consensus

First downstream ATG

Downstream ATGs in frame and highly translated don’t have a consensus

Except for 3rd-codon-position bias.

Calculate a wide and a narrow consensus sequence

Calculate motif score against the position weight matrix (pwm) for both narrow (-4 from ATG through to ATG) and wide (-10 from ATG to ATG) kozak consensus motif. These motifs are taken from the top 5% highly translated genes.

Estimate the information content

Using the sequence logo, details on https://en.wikipedia.org/wiki/Sequence_logo.

This is equal to the total height of the letters in the sequence logo summed across multiple positions.

## # A tibble: 6 x 4
##   Genes    ATG   Width  Infon
##   <chr>    <chr> <chr>  <dbl>
## 1 All      aATG  narrow 1.03 
## 2 HiTrans  aATG  narrow 2.84 
## 3 CytoRibo aATG  narrow 4.01 
## 4 All      d1ATG narrow 0.131
## 5 HiTrans  d1ATG narrow 0.307
## 6 CytoRibo d1ATG narrow 0.517

Information content in bits of highly-translated consensus (excluding 6 bits from ATG), narrow is 2.84, of wide is 3.81.

Estimate information content per base across start context

This is equal to the total height of the letters in the sequence logo at each position. It could be useful in comparing the total information without getting overly distracted by the actual letters.

Calculate scores of aATG, dATG, uATG against Kozak consensus

We calculate scores using Biostrings::PWMscoreStartingAt.

The best description I could find of this method is: https://support.bioconductor.org/p/61520/

It is just the sum of the matrix product of the PWM with the sequence.

Write scores to file scores_kozak_JEC21.txt.

## # A tibble: 6,636 x 13
##    Gene  aATG.scorekn d1.scorekn d2.scorekn u1.scorekn aATG.scorekw
##    <chr>        <dbl>      <dbl>      <dbl>      <dbl>        <dbl>
##  1 CNA0…        0.725      0.756      0.899      0.966        0.744
##  2 CNA0…        0.805      0.932      0.727      0.862        0.786
##  3 CNA0…        0.872      0.683      0.823      0.805        0.885
##  4 CNA0…        0.895      0.795      0.837     NA            0.858
##  5 CNA0…        0.964      0.741      0.839     NA            0.866
##  6 CNA0…        1.         0.884      0.698     NA            0.943
##  7 CNA0…        0.977      0.803      0.763     NA            0.919
##  8 CNA0…        0.781      0.964      0.758      0.733        0.769
##  9 CNA0…        0.940      0.920      0.803     NA            0.929
## 10 CNA0…        0.851      0.816      0.781     NA            0.862
## # ... with 6,626 more rows, and 7 more variables: d1.scorekw <dbl>,
## #   d2.scorekw <dbl>, u1.scorekw <dbl>, d1vsan <dbl>, u1vsan <dbl>,
## #   d1vsaw <dbl>, u1vsaw <dbl>

Plot against narrow consensus (-4 to ATG)

Narrow consensus comparing d1ATG to d2AUG score

Narrow consensus d1AUG score by frame

Plot against wide consensus (-10 to ATG)

Wide consensus comparing d1ATG to d2AUG score

For all annotated genes

For highly translated genes

For cytoplasmic ribosomal proteins

This illustrates that the MI between pairs of nts is generally weak. Except for the nts sharing a codon in the +4 to +12 positions. And secondarily at the -6 to -4 positions.

What is the -5 correlation in hiTrans genes?

What is the -5 correlation in cyto ribo genes?

So there is a strong tendency to have a C at -5 if there is a T at -6.

What about the -4 correlation in hiTrans genes?

What about the -4 correlation in cyto ribo genes?

This is very interesting: essentially everything with a -4A has a -2A and -1A, but a -4C is more relaxed.

What about the -1 correlation in hiTrans genes?

What about the -1 correlation in cyto ribo genes?

This looks interesting and needs a better way of summarizing.

Most dAUG scores are less than aAUG scores

Most u1AUG scores are less than aAUG scores

For highly translated genes, most dAUG scores are much less than aATG

Red: high dATG vs aATG Kozak score. Blue: highly translated. Purple: both.

dATG in frame with ATG, narrow

Files with high difference in narrow score, filtered for reasonable amounts of RNA, in frame. Saved to dvsaATG_highdiffn_inframe_JEC21.txt.

## # A tibble: 87 x 3
##    Gene     aATG.scorekn d1.scorekn
##    <chr>           <dbl>      <dbl>
##  1 CNA01530        0.673      1.   
##  2 CNI00670        0.698      1.   
##  3 CNI00690        0.673      0.945
##  4 CNA00760        0.733      1.   
##  5 CNB04570        0.730      0.989
##  6 CNF02520        0.673      0.925
##  7 CNH00360        0.740      0.989
##  8 CNG01740        0.742      0.989
##  9 CNN00820        0.699      0.945
## 10 CNB01775        0.756      1.   
## # ... with 77 more rows

dATG out of frame with ATG, narrow

Files with high difference in narrow score, filtered for reasonable amounts of RNA, out of frame. Saved to dvsaATG_highdiffn_outframe_JEC21.txt.

## # A tibble: 52 x 3
##    Gene     aATG.scorekn d1.scorekn
##    <chr>           <dbl>      <dbl>
##  1 CNI00340        0.706      0.966
##  2 CNC07140        0.761      1.   
##  3 CNN00160        0.759      0.989
##  4 CNB05380        0.758      0.977
##  5 CND06330        0.751      0.966
##  6 CND01650        0.707      0.922
##  7 CNG00490        0.792      1.   
##  8 CNK01150        0.761      0.952
##  9 CNC02170        0.811      1.   
## 10 CNE04530        0.811      1.   
## # ... with 42 more rows

dATG in frame with ATG, wide

Files with high difference in narrow score, filtered for reasonable amounts of RNA, in frame. Saved to dvsaATG_highdiffw_inframe_JEC21.txt.

## # A tibble: 86 x 3
##    Gene     aATG.scorekw d1.scorekw
##    <chr>           <dbl>      <dbl>
##  1 CNI00690        0.631      0.913
##  2 CNA04990        0.669      0.939
##  3 CNF02520        0.665      0.925
##  4 CNB01775        0.661      0.912
##  5 CNG01740        0.717      0.963
##  6 CNA05725        0.721      0.967
##  7 CNN00820        0.657      0.900
##  8 CNG01890        0.607      0.845
##  9 CNB04570        0.721      0.954
## 10 CNI00670        0.730      0.961
## # ... with 76 more rows

dATG out of frame with ATG, wide

Files with high difference in narrow score, filtered for reasonable amounts of RNA, out of frame. Saved to dvsaATG_highdiffw_outframe_JEC21.txt.

## # A tibble: 54 x 3
##    Gene     aATG.scorekw d1.scorekw
##    <chr>           <dbl>      <dbl>
##  1 CNI00340        0.665      0.953
##  2 CNC07140        0.664      0.912
##  3 CNE04530        0.757      0.980
##  4 CNN00160        0.721      0.930
##  5 CNN01750        0.759      0.953
##  6 CNG00500        0.736      0.928
##  7 CND01020        0.778      0.965
##  8 CNG02120        0.770      0.945
##  9 CND01650        0.727      0.902
## 10 CNF01580        0.724      0.898
## # ... with 44 more rows

dATG vs aATG ribosome occupancy depends on the context

For top 3315 / 50% of genes by mean RNA TPM.

dATG vs aATG ribosome occupancy depends on the context, geometric mean across reps

There is slight enrichment in high-score dATGs in frame near the aATG

Load predictions from mitofates

In input file JEC21_mitofates.txt.

Genes with high dATG vs aAUG score are enriched in mitochondrial presequences

There are spectacular correlations between d1 score, d1 frame, and localization

Count dATG vs AUG score, d1 frame, mito pre, enough RNA

## # A tibble: 16 x 5
## # Groups:   enoughR, d1vsaw0p1, d1.framefac [?]
##    enoughR d1vsaw0p1 d1.framefac Pred_preseq     n
##    <fct>   <fct>     <fct>       <fct>       <int>
##  1 Yes     d1lo      In          No            774
##  2 Yes     d1lo      In          Yes           123
##  3 Yes     d1lo      Out         No           2031
##  4 Yes     d1lo      Out         Yes           219
##  5 Yes     d1hi      In          No             52
##  6 Yes     d1hi      In          Yes            37
##  7 Yes     d1hi      Out         No             50
##  8 Yes     d1hi      Out         Yes            10
##  9 No      d1lo      In          No            826
## 10 No      d1lo      In          Yes            62
## 11 No      d1lo      Out         No           2125
## 12 No      d1lo      Out         Yes            89
## 13 No      d1hi      In          No            101
## 14 No      d1hi      In          Yes            13
## 15 No      d1hi      Out         No             96
## 16 No      d1hi      Out         Yes             6

However, mito-localized genes do not have a distinctive aATG context

It’s just a subset: the dual-localized ones.

Mitofates and d1ATG/d2ATG wide score

uATGs are associated with lower absolute translation

uATGs are associated with lower translation efficiency

uATGs associated with lower translation efficiency are over 20nt from TSS

uAUG score weakly affects TE

We suspect that uATG is associated with lower TE if the uATG has

• position at least 20nt downstream from TSS
• higher score

This figure shows that, for genes with only 1 uATG, this correlation is weak.

List of genes with low TE and uATGs far from TSS

Check these for ribosome occupancy at uATG.

## # A tibble: 0 x 8
## # Groups:   Gene [?]
## # ... with 8 variables: Gene <chr>, RNA <dbl>, RPF <dbl>, TE <dbl>,
## #   uATGCt <int>, uATGCtmin20 <int>, u1.cxtn <chr>, u2.cxtn <chr>

## # A tibble: 7 x 8
## # Groups:   Gene [?]
##   Gene       RNA   RPF     TE uATGCt uATGCtmin20 u1.cxtn  u2.cxtn 
##   <chr>    <dbl> <dbl>  <dbl>  <int>       <int> <chr>    <chr>   
## 1 CNA07610  50.0  5.85 0.117       1           1 TCCGTATG <NA>    
## 2 CNF00330 182.   3.20 0.0176      8           8 AAAAAATG CAAAAATG
## 3 CNG00290  58.3  4.38 0.0751      1           1 GCAGGATG <NA>    
## 4 CNG04240 123.   5.50 0.0446      0           0 <NA>     <NA>    
## 5 CNH02210  42.4  1.68 0.0396      1           1 CCACAATG <NA>    
## 6 CNL04930 203.  17.3  0.0853      2           2 CGACAATG ACTTTATG
## 7 CNM02470 171.  17.0  0.0993      2           2 CCAGAATG CCATCATG

uATG vs aATG ribosome occupancy depends on the context, narrow

For top 3315 / 50% of genes by mean RNA TPM.

uATG vs aATG ribosome occupancy depends on the context, wide all reps

For top 3315 / 50% of genes by mean RNA TPM, summarized by gene, both samples.

uATG vs aATG ribosome occupancy depends on the context, wide summarized

For top 3315 / 50% of genes by mean RNA TPM, with only a single uATG, summarized by gene, median across 4 samples.

Back to table of contents

Load expression data

## # A tibble: 6,790 x 4
## # Groups:   Gene [6,790]
##    Gene          RNA    RPF    TE
##    <chr>       <dbl>  <dbl> <dbl>
##  1 CNAG_06125 10270. 20140. 1.96 
##  2 CNAG_06101  8775.  8494. 0.968
##  3 CNAG_05762  7529.  7499. 0.996
##  4 CNAG_00779  3896.  7432. 1.91 
##  5 CNAG_03127  6254.  7164. 1.15 
##  6 CNAG_06222  6631.  6772. 1.02 
##  7 CNAG_04011 13306.  6772. 0.509
##  8 CNAG_01455 12670.  6548. 0.517
##  9 CNAG_05525  6970.  6461. 0.927
## 10 CNAG_03739  6515.  6383. 0.980
## # ... with 6,780 more rows

We also calculated hiTrans_H99, the top 5% (330) translated genes by RPF TPM.

Ribosome occupancy mostly tracks RNA abundance

Load context data

## # A tibble: 6,791 x 19
##    Gene  aATG.context aATG.pos d1.context d1.posTSS d1.posATG d1.frame
##    <chr> <chr>           <dbl> <chr>          <dbl>     <dbl>    <dbl>
##  1 CNAG… TACTTACGCGA…       70 AAATTCACT…       100        30        0
##  2 CNAG… GAACTTCGATC…       52 TCTCCCGCC…       114        62        2
##  3 CNAG… TGTCTCCTTGA…      104 ACTTACGCC…       189        85        1
##  4 CNAG… CACATACGTAA…      214 CCGAACGGC…       256        42        0
##  5 CNAG… GACTATACAAA…       55 GGAGGTGGG…       163       108        0
##  6 CNAG… AACCATACAAA…       99 CAAAGCCAT…       259       160        1
##  7 CNAG… ACCGTGCACAC…       75 GTATTCGGA…       105        30        0
##  8 CNAG… GTTTTCAACAG…       73 CCCATCAGA…       380       307        1
##  9 CNAG… GTACTATTGAA…      206 GAGGCTCCG…       513       307        1
## 10 CNAG… TACAAGCTTGA…       90 GGCCGCCTT…       151        61        1
## # ... with 6,781 more rows, and 12 more variables: d2.context <chr>,
## #   d2.posTSS <dbl>, d2.posATG <dbl>, d2.frame <dbl>, u1.context <chr>,
## #   u1.posTSS <dbl>, u1.posATG <dbl>, u1.frame <dbl>, u2.context <chr>,
## #   u2.posTSS <dbl>, u2.posATG <dbl>, u2.frame <dbl>

Annotated ATGs have a Kozak consensus sequence

Highly translated Annotated ATGs have a Kozak consensus sequence

That’s for hiTrans_H99, the top 5% (330) translated genes by RPF TPM.

Cytoplasmic Ribosome Annotated ATGs have a consensus sequence

Ideally would fix this more nicely.

All cytoribo genes are highly translated

Venn diagram

Upstream ATGs don’t have a consensus

First upstream ATG.

Downstream ATGs don’t have a consensus

First downstream ATG

Downstream ATGs in frame and highly translated don’t have a consensus

Except for 3rd-codon-position bias.

Calculate a wide and a narrow consensus sequence

Calculate motif score against the position weight matrix (pwm) for both narrow (-4 from ATG through to ATG) and wide (-10 from ATG to ATG) kozak consensus motif. These motifs are taken from the top 5% highly translated genes.

Estimate the information content

Using the sequence logo, details on https://en.wikipedia.org/wiki/Sequence_logo.

This is equal to the total height of the letters in the sequence logo summed across multiple positions.

## # A tibble: 6 x 4
##   Genes    ATG   Width  Infon
##   <chr>    <chr> <chr>  <dbl>
## 1 All      aATG  narrow 0.944
## 2 HiTrans  aATG  narrow 2.96 
## 3 CytoRibo aATG  narrow 4.16 
## 4 All      d1ATG narrow 0.116
## 5 HiTrans  d1ATG narrow 0.233
## 6 CytoRibo d1ATG narrow 0.492

Information content in bits of highly-translated consensus (excluding 6 bits from ATG), narrow is 2.96, of wide is 3.88.

Estimate information content per base across start context

Calculate scores of aATG, dATG, uATG against Kozak consensus

Write scores to file scores_kozak_H99.txt.

## # A tibble: 6,791 x 13
##    Gene  aATG.scorekn d1.scorekn d2.scorekn u1.scorekn aATG.scorekw
##    <chr>        <dbl>      <dbl>      <dbl>      <dbl>        <dbl>
##  1 CNAG…        0.904      0.776      0.747      0.847        0.849
##  2 CNAG…        0.871      0.841      1         NA            0.817
##  3 CNAG…        0.786      0.849      0.720     NA            0.769
##  4 CNAG…        0.920      0.785      0.788      0.849        0.866
##  5 CNAG…        0.978      0.759      0.822     NA            0.929
##  6 CNAG…        0.978      0.704      0.763     NA            0.928
##  7 CNAG…        0.966      0.856      0.849     NA            0.873
##  8 CNAG…        0.878      0.834      0.978     NA            0.851
##  9 CNAG…        0.891      0.787      0.821      0.821        0.900
## 10 CNAG…        0.874      0.935      0.662      0.678        0.825
## # ... with 6,781 more rows, and 7 more variables: d1.scorekw <dbl>,
## #   d2.scorekw <dbl>, u1.scorekw <dbl>, d1vsan <dbl>, u1vsan <dbl>,
## #   d1vsaw <dbl>, u1vsaw <dbl>

Plot against narrow consensus (-4 to ATG)

Plot against wide consensus (-10 to ATG)

Wide consensus comparing d1ATG to d2AUG score

For all annotated genes

For highly translated genes

For cytoplasmic ribosomal proteins

This illustrates that the MI between pairs of nts is generally weak. Except for the nts sharing a codon in the +4 to +12 positions. And secondarily at the -6 to -4 positions.

What is the -5 correlation in hiTrans genes?

What is the -5 correlation in cyto ribo genes?

So there is a strong tendency to have a C at -5 if there is a T at -6.

What about the -4 correlation in cyto ribo genes?

This is uninteresting because the counts are so low with no -4A. The -3A is even worse.

Most dAUG scores are less than aAUG scores

Most u1AUG scores are less than aAUG scores

For highly translated genes, most dAUG scores are much less than aATG

Filtered for enough RNA!

dATG in frame with ATG, narrow

Files with high difference in narrow score, filtered for reasonable amounts of RNA, in frame. Saved to dvsaATG_highdiffn_inframe_H99.txt.

## # A tibble: 97 x 3
##    Gene       aATG.scorekn d1.scorekn
##    <chr>             <dbl>      <dbl>
##  1 CNAG_04764        0.633      0.990
##  2 CNAG_04147        0.636      0.990
##  3 CNAG_00165        0.662      1    
##  4 CNAG_07473        0.633      0.968
##  5 CNAG_07801        0.662      0.978
##  6 CNAG_02259        0.662      0.978
##  7 CNAG_07776        0.688      1    
##  8 CNAG_04179        0.696      0.990
##  9 CNAG_01092        0.675      0.966
## 10 CNAG_00409        0.636      0.920
## # ... with 87 more rows

dATG out of frame with ATG, narrow

Files with high difference in narrow score, filtered for reasonable amounts of RNA, out of frame. Saved to dvsaATG_highdiffn_outframe_H99.txt.

## # A tibble: 52 x 3
##    Gene       aATG.scorekn d1.scorekn
##    <chr>             <dbl>      <dbl>
##  1 CNAG_06278        0.730      0.990
##  2 CNAG_04264        0.721      0.978
##  3 CNAG_04054        0.750      0.978
##  4 CNAG_03665        0.789      1    
##  5 CNAG_06627        0.770      0.978
##  6 CNAG_01999        0.733      0.938
##  7 CNAG_06006        0.747      0.938
##  8 CNAG_03811        0.790      0.978
##  9 CNAG_06344        0.752      0.935
## 10 CNAG_07979        0.818      1    
## # ... with 42 more rows

dATG in frame with ATG, wide

Files with high difference in narrow score, filtered for reasonable amounts of RNA, in frame. Saved to dvsaATG_highdiffw_inframe_H99.txt.

## # A tibble: 102 x 3
##    Gene       aATG.scorekw d1.scorekw
##    <chr>             <dbl>      <dbl>
##  1 CNAG_04147        0.611      0.929
##  2 CNAG_02259        0.641      0.953
##  3 CNAG_04764        0.616      0.905
##  4 CNAG_07801        0.682      0.945
##  5 CNAG_04179        0.691      0.947
##  6 CNAG_07776        0.712      0.967
##  7 CNAG_07473        0.639      0.893
##  8 CNAG_03953        0.687      0.940
##  9 CNAG_03486        0.700      0.951
## 10 CNAG_05722        0.652      0.901
## # ... with 92 more rows

dATG out of frame with ATG, wide

Files with high difference in narrow score, filtered for reasonable amounts of RNA, out of frame. Saved to dvsaATG_highdiffw_outframe_H99.txt.

## # A tibble: 44 x 3
##    Gene       aATG.scorekw d1.scorekw
##    <chr>             <dbl>      <dbl>
##  1 CNAG_06278        0.693      0.931
##  2 CNAG_03811        0.752      0.966
##  3 CNAG_03370        0.738      0.941
##  4 CNAG_06006        0.670      0.863
##  5 CNAG_06782        0.704      0.895
##  6 CNAG_01675        0.646      0.827
##  7 CNAG_06508        0.756      0.933
##  8 CNAG_07680        0.792      0.968
##  9 CNAG_00973        0.791      0.963
## 10 CNAG_01999        0.705      0.876
## # ... with 34 more rows

dATG vs aATG ribosome occupancy depends on the context

For top 3315 / 50% of genes by mean RNA TPM.

dATG vs aATG ribosome occupancy and narrow score, geometric mean across reps

dATG vs aATG ribosome occupancy and wide score, geometric mean across reps

Load predictions from mitofates

In input file H99_mitofates.txt.

Genes with high dATG vs aAUG score are enriched in mitochondrial presequences

There are spectacular correlations between d1 score, d1 frame, and localization

Count dATG vs AUG score, d1 frame, mito pre, enough RNA

## # A tibble: 16 x 5
## # Groups:   enoughR, d1vsaw0p1, d1.framefac [?]
##    enoughR d1vsaw0p1 d1.framefac Pred_preseq     n
##    <fct>   <fct>     <fct>       <fct>       <int>
##  1 Yes     d1lo      In          No            752
##  2 Yes     d1lo      In          Yes           120
##  3 Yes     d1lo      Out         No           2018
##  4 Yes     d1lo      Out         Yes           241
##  5 Yes     d1hi      In          No             69
##  6 Yes     d1hi      In          Yes            41
##  7 Yes     d1hi      Out         No             53
##  8 Yes     d1hi      Out         Yes             4
##  9 No      d1lo      In          No            906
## 10 No      d1lo      In          Yes            55
## 11 No      d1lo      Out         No           2222
## 12 No      d1lo      Out         Yes            81
## 13 No      d1hi      In          No             98
## 14 No      d1hi      In          Yes            10
## 15 No      d1hi      Out         No             91
## 16 No      d1hi      Out         Yes             7

However, mito-localized genes do not have a distinctive aATG context

It’s just a subset: the dual-localized ones.

Load predictions from DeepLoc

In input file H99_DeepLoc.txt.

Compare mitofates and DeepLoc mitochondrial predictions

Compare mitofates and DeepLoc plastid predictions

There are correlations between d1 score, d1 frame, and DeepLoc predictions

Load predictions

In input file H99_SignalP.txt.

## # A tibble: 15 x 5
## # Groups:   enoughRNA, d1vsawfac, d1.framefac [?]
##    enoughRNA  d1vsawfac                d1.framefac SignalP     n
##    <fct>      <fct>                    <fct>       <chr>   <int>
##  1 enough RNA "AUG score\nd < a + 0.1" In          N         836
##  2 enough RNA "AUG score\nd < a + 0.1" In          Y          36
##  3 enough RNA "AUG score\nd < a + 0.1" Out         N        2148
##  4 enough RNA "AUG score\nd < a + 0.1" Out         Y         111
##  5 enough RNA "AUG score\na + 0.1 < d" In          N         107
##  6 enough RNA "AUG score\na + 0.1 < d" In          Y           3
##  7 enough RNA "AUG score\na + 0.1 < d" Out         N          57
##  8 not enough "AUG score\nd < a + 0.1" In          N         909
##  9 not enough "AUG score\nd < a + 0.1" In          Y          52
## 10 not enough "AUG score\nd < a + 0.1" Out         N        2163
## 11 not enough "AUG score\nd < a + 0.1" Out         Y         141
## 12 not enough "AUG score\na + 0.1 < d" In          N         105
## 13 not enough "AUG score\na + 0.1 < d" In          Y           3
## 14 not enough "AUG score\na + 0.1 < d" Out         N          96
## 15 not enough "AUG score\na + 0.1 < d" Out         Y           2

Compare SignalP and DeepLoc plastid predictions

There are correlations between d1 score, d1 frame, and SignalP predictions

uATGs are associated with lower absolute translation

uATGs are associated with lower translation efficiency

uATGs associated with lower translation efficiency are over 20nt from TSS

uAUG score weakly affects TE

We suspect that uATG is associated with lower TE if the uATG has

• position at least 20nt downstream from TSS
• higher score

This figure shows that, for genes with only 1 uATG, this correlation is weak.

List of genes with low TE and uATGs far from TSS

Many of these (CNAG_03140, CNAG_07695, CNAG_06246) are strongly translationally repressed and have good context at the uATG.

## # A tibble: 0 x 8
## # Groups:   Gene [?]
## # ... with 8 variables: Gene <chr>, RNA <dbl>, RPF <dbl>, TE <dbl>,
## #   uATGCt <int>, uATGCtmin20 <int>, u1.cxtn <chr>, u2.cxtn <chr>

## # A tibble: 5 x 8
## # Groups:   Gene [?]
##   Gene         RNA   RPF     TE uATGCt uATGCtmin20 u1.cxtn  u2.cxtn 
##   <chr>      <dbl> <dbl>  <dbl>  <int>       <int> <chr>    <chr>   
## 1 CNAG_00784  52.8  6.69 0.127       1           0 TCCGTATG <NA>    
## 2 CNAG_03578  43.5  6.20 0.143       1           1 GCAGGATG <NA>    
## 3 CNAG_05574  30.0  2.66 0.0885      1           1 CCACAATG <NA>    
## 4 CNAG_06246 196.  24.6  0.125       2           2 CCAGAATG CCATCATG
## 5 CNAG_07813 148.  20.1  0.136       1           1 CGGCAATG <NA>

uATG vs aATG ribosome occupancy depends on the context, narrow

For top 3315 / 50% of genes by mean RNA TPM.

uATG vs aATG ribosome occupancy depends on the context, wide all reps

For top 3315 / 50% of genes by mean RNA TPM, summarized by gene, all 4 samples.

uATG vs aATG ribosome occupancy depends on the context, wide summarized

For top 3315 / 50% of genes by mean RNA TPM, with only a single uATG, summarized by gene, median across 4 samples.

Back to table of contents

RPF vs RNA with same colours

RNA Abundance

Ribosome Occupancy

Translation efficiency, no threshold

Translation efficiency, filtered by top 50% of expression

dATG in frame with ATG

Saved to file dvsaATG_highdiffn_inframe_cc.txt.

## # A tibble: 47 x 6
##    H99        JEC21    a.skn.H99 d.skn.H99 a.skn.JEC21 d.skn.JEC21
##    <chr>      <chr>        <dbl>     <dbl>       <dbl>       <dbl>
##  1 CNAG_00165 CNA01530     0.662     1           0.673       1    
##  2 CNAG_07776 CNI00670     0.688     1           0.698       1    
##  3 CNAG_07473 CNB01880     0.633     0.968       0.655       0.888
##  4 CNAG_07801 CNL06190     0.662     0.978       0.673       0.899
##  5 CNAG_00086 CNA00760     0.725     1           0.733       1    
##  6 CNAG_04179 CNI03160     0.696     0.990       0.715       0.945
##  7 CNAG_07873 CNH00360     0.721     0.990       0.740       0.989
##  8 CNAG_05722 CNF02520     0.662     0.916       0.673       0.925
##  9 CNAG_06353 CNN00820     0.688     0.945       0.699       0.945
## 10 CNAG_03953 CNB04410     0.723     0.990       0.741       0.945
## 11 CNAG_02306 CNE02400     0.723     0.968       0.741       0.966
## 12 CNAG_02545 CNE00210     0.696     0.938       0.715       0.940
## 13 CNAG_02259 CNE02870     0.662     0.978       0.683       0.827
## 14 CNAG_01544 CNC06400     0.759     0.978       0.756       0.977
## 15 CNAG_03996 CNB04810     0.696     0.895       0.655       0.886
## 16 CNAG_04219 CNI03610     0.770     0.978       0.761       0.977
## 17 CNAG_02880 CNC04930     0.696     0.916       0.715       0.917
## 18 CNAG_03679 CNB01775     0.822     1           0.756       1    
## 19 CNAG_02431 CNE01260     0.759     1           0.763       0.940
## 20 CNAG_00517 CNA04990     0.728     0.938       0.736       0.940
## # ... with 27 more rows

dATG out of frame

Saved to file dvsaATG_highdiffn_outframe_cc.txt.

## # A tibble: 14 x 6
##    H99        JEC21    a.skn.H99 d.skn.H99 a.skn.JEC21 d.skn.JEC21
##    <chr>      <chr>        <dbl>     <dbl>       <dbl>       <dbl>
##  1 CNAG_06278 CNN00160     0.730     0.990       0.759       0.989
##  2 CNAG_04054 CNB05380     0.750     0.978       0.758       0.977
##  3 CNAG_02809 CNC04270     0.800     0.978       0.793       0.977
##  4 CNAG_03008 CNC06190     0.827     1           0.819       1    
##  5 CNAG_03370 CNG02120     0.795     0.968       0.786       0.966
##  6 CNAG_01667 CNC01780     0.835     0.990       0.828       0.989
##  7 CNAG_03190 CNG03790     0.812     0.966       0.809       0.964
##  8 CNAG_04896 CNJ03200     0.821     0.975       0.823       0.975
##  9 CNAG_02894 CNC05065     0.819     0.978       0.837       0.977
## 10 CNAG_02578 CNK00690     0.818     0.944       0.785       0.941
## 11 CNAG_01241 CND03590     0.747     0.878       0.749       0.895
## 12 CNAG_06899 CNK00290     0.836     0.975       0.845       0.975
## 13 CNAG_07780 CNI00090     0.871     1           0.861       1    
## 14 CNAG_03839 CNB03280     0.817     0.938       0.813       0.940

• CNN00160 two-component-like sensor kinase TCO7
• CNB05380 SUI1/eIF1, translation initiation factor.
• CNC06190 has a domain conserved with eIF2.
• CNG02120 freqenin calcium-binding protein, FRQ1 homolog
• CNC05065/CNM00150/CNC04270/CNC06190/CNA07610/CND03900, all hypothetical or uncharacterized.
• CNC01780 prenyltransferase, COQ2 homolog
• CNK00690 regulation of meiosis-related, PCH2 homolog
• CND03590 protein phosphatase I nuclear regulatory subunit, SDS22 homolog
• CNI00090 farnesyltranstransferase, BTS1 homolog
• CND03900 has a W2 eIF4-gamma/eIF5/eIF2-epsilon - like domain
• CNB03280 mRNA transcription modulator, CCR4-NOT ubiquitin ligase subunit MOT2 homolog.

Genes with very different aAUG scores, JEC21 vs H99

Filtered for enough RNA (top 50%)

## # A tibble: 135 x 6
##    H99        JEC21    a.skn.H99 d.skn.H99 a.skn.JEC21 d.skn.JEC21
##    <chr>      <chr>        <dbl>     <dbl>       <dbl>       <dbl>
##  1 CNAG_04147 CNI02850     0.636     0.990       0.989       0.884
##  2 CNAG_01092 CND02180     0.675     0.966       0.964       0.917
##  3 CNAG_03486 CNG01060     0.691     0.966       0.964       0.813
##  4 CNAG_06196 CNM01950     0.721     0.662       0.989       0.733
##  5 CNAG_00690 CNA06680     0.704     0.895       0.964       0.683
##  6 CNAG_01188 CND03130     0.636     0.828       0.888       0.733
##  7 CNAG_06000 CNM00090     0.717     0.911       0.964       0.745
##  8 CNAG_07437 CND03020     0.720     0.966       0.964       0.888
##  9 CNAG_05678 CNF02960     0.675     0.906       0.914       0.929
## 10 CNAG_06446 CNN01710     0.762     1           1           0.658
## # ... with 125 more rows

## # A tibble: 135 x 6
##    H99        JEC21    a.skn.H99 d.skn.H99 a.skn.JEC21 d.skn.JEC21
##    <chr>      <chr>        <dbl>     <dbl>       <dbl>       <dbl>
##  1 CNAG_04488 CNI00690     0.945     0.968       0.673       0.945
##  2 CNAG_03968 CNB04570     0.990     0.757       0.730       0.989
##  3 CNAG_00529 CNA05110     0.953     0.803       0.698       0.892
##  4 CNAG_03410 CNG01740     0.990     0.730       0.742       0.989
##  5 CNAG_05720 CNF02540     1         0.746       0.756       1    
##  6 CNAG_04751 CNJ01820     0.966     0.854       0.725       0.964
##  7 CNAG_04899 CNJ03230     0.953     0.819       0.723       0.966
##  8 CNAG_04089 CNB05680     0.966     0.809       0.743       0.886
##  9 CNAG_05504 CNH01580     0.966     0.812       0.776       0.842
## 10 CNAG_05692 CNF02820     0.919     0.783       0.732       0.845
## # ... with 125 more rows

Many of these have the expected structure where homologs differ only at the N-terminus. There appears to be a swap between a near-ATG start codon, and a poor-context ATG, between the species.

Higher aAUG score in JEC21:

• CNAG_04147/CNI02850, putative RNA helicase RRP3 homolog
• CNAG_06000/CNM00090, putative glycoprotein
• CNAG_03486/CNG01060, peptidylprolyl isomerase, Scer has CPR2/CPR5 homolgs localized to ER and vacuole.
• CNAG_06353/CNN00820, tRNA pseudouridine synthase, exciting, CNAG_06353 has weak mito loc seq, Scer has PUS1/PUS2 nuc/mito homologs.
• CNAG_06446/CNN01710, mitochondrial splicing suppressor Mss51 homolog, clear mito preseq in both.
• CNAG_01092/CND02180, hypothetical protein conserved in crypto, mito preseq in both
• CNAG_01188/CND03130, ATGATG start in H99, not interesting
• CNAG_05678/CNF02960, putative transmembrane protein involved in ammonia production.
• CNAG_07645/CNE03115 Autophagy protein Atg12, H99 CNAG_07645 has non-homologous annotated N-terminal extension that seems improbably.

Higher aAUG score in H99:

• CNAG_00529/CNA05110, sulfite transporter, Scer Ssu1 homolog.
• CNAG_03410/CNG01740, upstream ATG in JEC21 close to TSS, prob not real
• CNAG_04089/CNB05680, again upstream ATG in JEC21 close to TSS, prob not real. Note that CNAG_03410/CNAG_04089 are paralogs
• CNAG_04751/CNJ01820, hypothetical protein conserved in tremellomycetes.
• CNAG_02703/CNK01910, hypothetical protein conserved in tremellomycetes.
• CNAG_03638/CNB01390, identical ATG context in fungidb, check.
• CNAG_05504/CNH01580, Iah1, pattern unclear, JEC21 just has poor start codons.
• CNAG_05692/CNF02820, sphinganine kinase, Scer YSR3/LCB2 homolog.
• CNAG_07609/CNC03180, putative RNA helicase, weak start context in JEC21 predicts less protein in cell. Or most translation initiation is from much later ATG? DDX51/DBP6 homolog.

These look like mostly misannotated in one strain, or not interesting. Is the upstream start codon in one strain actually used? Check for ribosome footprints and for other features (homology, mito localization seq). It would be nice to have an additional filter here.

Non-canonical aATG motif from the 9 tRNA synthetases with poor Kozak score and in-frame high-score dATG

dATG motif is canonical from those 9 tRNA synthetases

Calculate unusual tRNA synthetase scores and consensus table

Output to file scores_kozak_unusual_mito_cc.txt.

## # A tibble: 6,341 x 14
##    H99   JEC21 aATG.scorekn.H99 d1.scorekn.H99 u1.scorekn.H99
##    <chr> <chr>            <dbl>          <dbl>          <dbl>
##  1 CNAG… CND0…            0.990          0.785         NA    
##  2 CNAG… CNH0…            0.873          0.922          0.850
##  3 CNAG… CNH0…            0.873          0.850          0.880
##  4 CNAG… CNB0…            0.905          0.783          0.795
##  5 CNAG… CNF0…            0.841          0.721         NA    
##  6 CNAG… CNA0…            1              1              0.953
##  7 CNAG… CNK0…            0.883          0.776         NA    
##  8 CNAG… CNB0…            0.966          0.721         NA    
##  9 CNAG… CNM0…            0.945          0.741         NA    
## 10 CNAG… CNG0…            0.953          0.893          0.805
## # ... with 6,331 more rows, and 9 more variables:
## #   aATG.scorekn.JEC21 <dbl>, d1.scorekn.JEC21 <dbl>,
## #   u1.scorekn.JEC21 <dbl>, aATG.scoreUn.H99 <dbl>,
## #   aATG.scoreUn.JEC21 <dbl>, Prob_mito.H99 <dbl>, Pred_mito.H99 <fct>,
## #   Prob_mito.JEC21 <dbl>, Pred_mito.JEC21 <fct>

Genes with high score against utRS motif are often dual-localized.

## # A tibble: 21 x 6
##    Gene     a.scorekn d.scorekn a.scoreUn aATG.context         Prob_preseq
##    <chr>        <dbl>     <dbl>     <dbl> <chr>                      <dbl>
##  1 CNA01530     0.673     1         0.982 AACAAACATTGTATGCGTT…       0.299
##  2 CNB01880     0.655     0.888     0.982 CGATAACGGTGTATGCTCC…       0.967
##  3 CNL06190     0.673     0.899     0.982 ATCTTCGTTTGTATGCCAA…       0.443
##  4 CNA00760     0.733     1         0.972 AAACATCTTTGCATGCCTC…       0.546
##  5 CNI03160     0.715     0.945     0.972 TCCAGATTGTGCATGCTCA…       0.985
##  6 CNF02520     0.673     0.925     0.982 ACACGACATTGTATGCTCT…       0.989
##  7 CNE00210     0.715     0.940     0.972 AAAGAAAGGTGCATGAACA…       0.9  
##  8 CNC06400     0.756     0.977     0.951 CCCAGCAGCTTTATGCTCC…       0.995
##  9 CNB04810     0.655     0.886     0.982 ATCAACAGGTGTATGCCTG…       0    
## 10 CNC04930     0.715     0.917     0.972 GAAAAGAGGTGCATGTCAT…       0.507
## # ... with 11 more rows

These genes are:

- CNB01880, valine tRNA-ligase*
- CNA01530, S-methyl-5-thioadenosine phosphorylase MEU1 homolog
- CNE00210, inorganic diphosphatase IPP1 homolog
- CNL06190, putative aldo-keto reductase GCY1/YCR1 homolog
- CNC06400, histidine-tRNA ligase*
- CNI03160, lysine-tRNA ligase*
- CNF02520, alanine-tRNA ligase*
- CNC04930, YchF-family translational regulator, OLA1 homolog
- CNB04810, guanosine-diphosphatase, GDA1 homolog
- CNG01890, NAD+ diphosphatase, NPY1 homolog
- CNA00760, tRNA (guanine-N2-)-methyltransferase, TRM1 homolog
- CNJ00430, tryptophan-tRNA ligase*
- CNB00900, cysteine-tRNA ligase*
- CNB05640, proline-tRNA ligase*
- CNB01800, macrolide-binding protein FKBP12, FPR1 homolog
- CNB04990, Methionine-R-sulfoxide reductase, MXR2 homolog
- CNB01775, hypothetical Acylphosphatase family protein
- CNB04910, putative aminopeptidase, AAP1/APE2 homolog
- CND00900, DEAH-box RNA helicase, PRP22 homolog

* these genes were used to construct the consensus sequence.

In S. cerevisiae: TRM1 is dual-localized to mito and cyto. APE2 is dual mito/cyto localized. GDA1 is golgi-localized. NPY1 is peroxisome and cyto localized. GCY1 and YCR1 are nucleus/cytoplasm localized. FPR1 is multiply localized. MXR2 is mito localized.

CND00900 has two adjacent ATGs to start (GCTTTTATGATGTCC), the 2nd of which gives the good context score due to the A in the first.

aATG motif scores, comparing Kozak vs non-canonical tRS

In JEC21 only. Red: high dATG vs aATG Kozak score. Blue: highly translated. Purple: both.

Note that many points are overplotted at the same y-values because the score is mostly quantized on a small number of values.

High utRS score is slightly enriched in mito-localized genes

Note we do not control for expression or any other potential factor affecting ATG context.

To-do: Analysis of other tRNA synthetases

For example, CNM01290, Leucine-tRNA synthetase, was on a previous version of this list. It has good kozak score at aATG and conserved out-of-frame dATG.

dAUG score high compared aATG and dATG out of frame, 14 genes

File dvsaATG_highdiffn_outframe_cc.txt.

No significant GO terms.

dAUG score high compared aATG and dATG in frame, 44 genes

File dvsaATG_highdiffn_inframe_cc.txt.

Enriched in Biological processes:

• tRNA aminoacylation for protein translation< translation < protein metabolic process < primary metabolic process < metabolic process
• tRNA metabolic process < RNA metabolic process < nucleobase-containing compound metabolic process
• Unclassified

Molecular Function:

• aminoacyl-tRNA ligase activity < ligase activity < catalytic activity

Cellular Component:

• cytosol < cytoplasm < intracellular < cell part

Highly translated, 291 genes

File hiTrans_cc.txt.

Enriched BPs include:

• oxidative phosphorylation
• glycolysis
• translation
• protein folding
• cation transport
• mitochondrial transport

Enriched MFs include:

• transmembrane transporter activity
• structural constituent of ribosome
• translation elongation factor activity

Enriched CCs include:

• proton-transporting ATP synthase complex
• ribosome
• mitochondrial inner membrane

High translation efficiency, 174 genes

File hiTE_cc.txt.

Enriched BPs include:

• pentose-phosphate shunt < monosaccharide metabolic process
• tRNA aminoacylation for protein translation < translation
• acyl-CoA metabolic process
• tricarboxylic acid cycle
• cellular amino acid biosynthetic process
• nuclear transport

Enriched MFs include:

• translation elongation factor activity
• translation initiation factor activity
• aminoacyl-tRNA ligase activity

Enriched CCs include:

• cytosol < cytoplasm

Low translation efficiency, 284 genes

File loTE_cc.txt.

Enriched BPs include:

• anion transport

Enriched MFs, no sig. results.

Enriched CCs, no sig. results.

Back to table of contents

Figure 2 panels: TE and uAUG usage in H99

Supp fig2: TE and uAUG usage in JEC21

Figure 3 panels: score and AUG/uAUG usage in H99

Figure S3 panels: score and AUG/uAUG usage in JEC21

Figure 6 top: score and AUG/dAUG usage in H99